[PATCH V2 11/15] cpufreq: mediatek: Update logic of voltage_tracking()

Rex-BC Chen rex-bc.chen at mediatek.com
Thu Apr 14 04:30:20 PDT 2022


On Fri, 2022-04-08 at 14:08 -0700, Kevin Hilman wrote:
> Rex-BC Chen <rex-bc.chen at mediatek.com> writes:
> 
> > From: Jia-Wei Chang <jia-wei.chang at mediatek.com>
> > 
> > - Remove VOLT_TOL because CCI may share the same sram and vproc
> >   regulators with CPU. Therefore, set the max voltage in
> >   regulator_set_voltage() to the proc{sram}_max_volt.
> 
> This could you a bit more detailed explanation.  Why does VOLT_TOL
> get
> in the way when regulators are shared between CPU & CCI?

Hello Kevin,

Here we use 'sram_min_volt' and 'sram_max_volt' to determine the
voltage boundary of sram regulator.
And use (sram_min_volt - min_volt_shift) and 'proc_max_volt' to
determine the voltage boundary of vproc regulator.
We use them as platform data to replace VOLT_TOL (voltage tolerance)
when determing the voltage boundary and invoking regulator_set_voltage.

I will add this to commit message in next version.

> 
> > - Move comparison of new and old voltages to
> >   mtk_cpufreq_voltage_tracking().
> 
> Why?  And how is this related to the above change?  Seems to me that
> it
> belongs in a separate patch.
> 
> Kevin

I think there are some mistake for this.
I will remove this commit message in next version.

BRs,
Rex




More information about the Linux-mediatek mailing list